If you're scanning Netflix for something worthwhile tonight, consider , a new documentary added on Jan. 1. The movie goes behind the scenes of artificial intelligence's landmark victory over a human champion, and more profoundly, hints at what the AI-driven future really looks like.
In March 2016, an AI by Google's DeepMind took on South Korea's Lee Sedol, perhaps the world's greatest flesh-and-blood Go player. Sedol reportedly expected a swift and decisive victory over the machines. And why not? While computers solved checkers and pummeled Homo sapiens at chess, mastery of Go had eluded them. The complexity of the ancient game seemed too much to overcome with the brute force of even modern supercomputing, as Go contains more possible positions than the number of atoms in the universe.
It would not play out this way. AlphaGo crushes Sedol, winning every match but one. The much-lauded movie captures the programmers' elation and the player's disappointment as the "Roger Federer of Go" goes down in flames.
What's more interesting than technology's inevitable triumph is the way that it happened. The original AlphaGo program brushed up on human games and human strategies as its method of learning the game and then extrapolated from there, combining that knowledge with its amazing computing power to unveil moves and gambits that no human player would've invented.
But in 2017, a year after Sedol's defeat, DeepMind one-upped itself. It released , a new Go-playing program that pummeled the original AlphaGo by embracing a new tactic. Instead of studying humanity's Go playbook, AlphaGo Zero learned the game from scratch by playing against itself. It didn't even need human help to become the best player ever at a human game.
As PM has noted before, these machines that learn without us are something wild. Maybe they're even the first real form of alien intelligence that we'll encounter. As the movie says, "These ideas that are driving AlphaGo are going to drive our future."